F1000Research · 2026
Dataset of multi-focus (Z-stack) images derived from liquid-based cervical cancer cytology specimens.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Onishi Takafumi, Miyamoto Tomoyuki, Osawa Yukihiko, Shibahara Kazuki, Nishimori Makoto, Yakushiji Hiromasa, Ohno Setsuyo, Ohno Eiji
Аннотация
BACKGROUND: In cervical cancer screening, cytotechnologists and cytopathologists integrate three-dimensional information by continuously adjusting the microscope's focus to evaluate chromatin structure and nuclear morphology. However, most existing public datasets consist of single-focus 2D images, which do not fully reflect this clinical diagnostic workflow. This study presents the Cervical Cancer Cell Image Database: Multi-focus Cytology Dataset (CCCID) to bridge this gap.
METHODS: Cervical specimens were processed using the BD SurePath™ LBC technique and Papanicolaou staining. Digitization was performed using a NanoZoomer-XR scanner. For 639 unique fields of view (FOVs), a Z-stack consisting of 11 focal planes was captured at 1.0 μm intervals, resulting in 7,029 images (384 × 384 pixels). Ground-truth labels were established only when six board-certified expert cytotechnologists reached 100% consensus.
CONCLUSIONS: The CCCID provides a high-reliability benchmark for developing machine-learning models that utilize axial (Z-axis) information. It is highly valuable for advancing three-dimensional nuclear morphology analysis, cell segmentation in overlapping clusters, and the evaluation of focus-fusion algorithms in digital cytopathology.
METHODS: Cervical specimens were processed using the BD SurePath™ LBC technique and Papanicolaou staining. Digitization was performed using a NanoZoomer-XR scanner. For 639 unique fields of view (FOVs), a Z-stack consisting of 11 focal planes was captured at 1.0 μm intervals, resulting in 7,029 images (384 × 384 pixels). Ground-truth labels were established only when six board-certified expert cytotechnologists reached 100% consensus.
CONCLUSIONS: The CCCID provides a high-reliability benchmark for developing machine-learning models that utilize axial (Z-axis) information. It is highly valuable for advancing three-dimensional nuclear morphology analysis, cell segmentation in overlapping clusters, and the evaluation of focus-fusion algorithms in digital cytopathology.